环境安装3.11语音模型

M1 Mac Anaconda

curl -O https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-arm64.sh
sh Miniconda3-latest-MacOSX-arm64.sh

不自动激活 conda config --set auto*activate*base false

撤销设置 conda init --reverse $SHELL

创建激活Conda环境

conda create -n llm_audio python=3.9
conda activate llm_audio

查看环境

conda info --envs | grep "*"

删除环境

conda remove --name llm_audio --all

安装 modelscope

audio,cv 版本不兼容需要单独环境安装

# audio,cv == 1.25.0

pip install "modelscope[audio]" -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html

pip install "modelscope[cv]" -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html

modelscope

Paraformer语音识别-中文-通用-16k-离线-large-pytorch

from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks

inference_16k_pipline = pipeline(
    task=Tasks.auto_speech_recognition,
    model='iic/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch')

rec_result = inference_16k_pipline('https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav')
print(rec_result)

语音合成-中文-多情感领域-16k-多发音人

from modelscope.outputs import OutputKeys
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks

text = '待合成文本'
model_id = 'iic/speech_sambert-hifigan_tts_zh-cn_16k'
sambert_hifigan_tts = pipeline(task=Tasks.text_to_speech, model=model_id)
output = sambert_hifigan_tts(input=text, voice='zhitian_emo')
wav = output[OutputKeys.OUTPUT_WAV]
with open('output.wav', 'wb') as f:
    f.write(wav)

Huggingface

Issues

ImportError: cannot import name 'kaiser' from 'scipy.signal'

pip install scipy==1.10.0

'ttsfrd' has no attribute 'TtsFrontendEngine'

raise type(e)(f'{obj_cls.name}: {e}') from e
AttributeError: TextToSpeechSambertHifiganPipeline: SambertHifigan: module 'ttsfrd' has no attribute 'TtsFrontendEngine'

CosyVoice

ttsfrd-0.0.3-cp39-cp39-macosx_12_0_x86_64.whl